Editorial
eBook
Web Marketing Moneyball: How Web Operations Drive Marketing ROI
The Gist
- AI is improving visibility faster than execution. Marketing teams can see churn risk, engagement decay and customer friction sooner, but many still lack a defined path for what happens next.
- Dashboards are becoming spectator tools. A signal has no value unless it changes a decision, assigns ownership and creates measurable action.
- The AI plateau is structural unreadiness. The issue is not weak models. It is an operating model that was never built to convert intelligence into action.
- Decision design is the fix. Define the signal, decision, owner and measure before adding more AI to the stack.
A marketing team gathers around a dashboard and watches a customer leave. Not literally, of course. Nobody sees the customer close the browser, ignore the next renewal reminder, skip the replenishment email and drift into silence. The team sees cleaner signals than that. A churn score rises, engagement drops, a segment turns red and someone says, “We should keep an eye on that.”
That sentence is where AI value goes to die. The model worked, the dashboard worked, the CRM worked and the team had more visibility than it had five years ago. Still, nothing changed because no one had designed the next decision. The system detected risk, but the organization treated detection as action. That is not intelligence, it is dashboard spectatorship.
This is the real AI plateau in marketing, CRM and martech. It is not that AI cannot see the customer clearly enough. It is that many organizations still have no operating model for what happens after the system sees something. More capability, less meaning. More prediction, less ownership.
Chris Willis, chief design officer and futurist at Domo, framed the problem as a foundation issue, not a tooling issue.
“AI transformation relies on solid foundations beyond data and tooling,” Willis told me. “Organizations need to focus on their why, how their people best operate, and the processes that have provided historical differentiation.” That is the missing piece in many AI strategies. The issue is not access to AI. It is structural unreadiness.
Table of Contents
- AI Did Not Create the Execution Gap
- Why Dashboards Don't Create Decision Ownership
- How AI Amplifies Broken Operating Models
- Marketing Keeps Buying Capability it Has Not Operationalized
- Real Time Can Make the Wrong System Worse
- Decision Design Turns Signals Into Action
- AI Decision-Making FAQ
- How to Convert AI Signals Into Action
AI Did Not Create the Execution Gap
Marketing had an execution gap long before AI entered the room. Teams already had dashboards, attribution models, customer profiles, lead scores, journey maps and campaign reports. Each layer promised better insight, but many of those insights never changed the next decision. AI did not create that failure. It put it on a bigger screen.
The data supports that diagnosis. In McKinsey’s 2025 State of AI survey, 88% of organizations reported using AI in at least one business function, but only 39% reported enterprise-wide EBIT impact. McKinsey also found that only about 6% of respondents qualified as AI high performers. Those high performers were more likely to redesign workflows and establish senior ownership of AI decisions.
The Gap Is in What Happens After Deployment
That distinction matters. The gap is not between companies using AI and companies avoiding it. The gap is between companies deploying AI and companies redesigning how work happens after AI produces an output. AI adoption is easy to count. AI impact is harder because it requires a change in operational reality.
BCG’s research points in the same direction. In BCG’s analysis of AI value generation, only 5% of companies qualify as future built, while 60% report minimal revenue or cost gains despite substantial AI investment. That is not a software access problem. It is a work design problem.
This is where the AI conversation keeps drifting in the wrong direction. Leaders keep asking whether the model is strong enough, whether the data is deep enough and whether the platform is advanced enough. Those are valid questions, but they are incomplete. The better question is whether the organization has defined what decision must change when the AI output appears.
Related Article: Why Marketing — and Not IT — Must Lead the AI Transformation
Why Dashboards Don't Create Decision Ownership
Dashboards are useful until teams start treating them like the work. They show movement, risk, leakage and friction. They help teams spot patterns faster. They do not create ownership, governance or follow-through by themselves.
Willis made that distinction clear.
“Dashboards are always a step in a much bigger process,” he said. “The insights tend to stall when the bigger process is not clearly defined or is misunderstood.” That is the operating gap hiding behind many AI programs. The data flow exists, the model output exists and the visualization exists. The decision path does not.
A check-engine light does not fix the car. It tells the driver that something requires attention. If the driver keeps driving until smoke comes from the hood, the problem is not the dashboard light. The problem is the lack of response. A lot of AI in marketing has become a very expensive check-engine light.
More Data Doesn't Mean Better Decisions
Oracle’s Decision Dilemma study shows why this matters. The study, conducted with more than 14,000 employees and business leaders across 17 countries, found that 86% said the volume of data made decisions in their personal and professional lives more complicated. It also found that 70% had given up on making a decision because the data felt overwhelming. More information did not create more confidence.
That should concern every marketing leader investing in AI. If the organization already struggles to make decisions from existing data, adding AI creates more signal for a weak decision system to process. The result is not clarity. The result is more dashboards, more alerts, more debate and the same slow response.
How AI Amplifies Broken Operating Models
AI magnifies the system it enters. In a disciplined operating model, it can create leverage. In a fragmented operating model, it creates louder noise, faster motion and more visible confusion. Tool-first thinking is the problem, not the solution.
Willis put it directly.
“Advancements in capabilities often magnify the good or bad in an organization,” he said. “Putting better systems in place will not automatically make a company better. In fact, sometimes they make a company worse or more chaotic.” That is the AI reality check marketing needs. No platform can unify what the operating model has not prepared to unify.
Blue Ridge Partners makes the same point in its commercial AI ROI research. The firm found that 71% of commercial leaders view AI as necessary, but say proving ROI remains a major challenge. More than half said AI investment decisions were driven more by fear of missing out than by strategy. That is not transformation. It is procurement pressure dressed as progress.
Where the Breakdown Shows Up in Daily Operations
In practice, the breakdown is easy to spot. A model recommends a priority account, but the sales team uses a different list. A health score flags retention risk, but the account team lacks a defined intervention. A journey platform predicts the next best action, but no one defines when a customer leaves one path and enters another. The technology functions, while the system fails.
That is why AI programs often look productive in demos and disappointing in daily operations. The demo shows the output. The workflow reveals the breakdown. The issue is not the model’s intelligence. It is the organization’s failure to embed ownership where execution lives.
Related Article: Dear CMOs: Your Problem Isn't Your AI. It's Your Operating Model.
Marketing Keeps Buying Capability it Has Not Operationalized
Martech has trained teams to treat more capability as progress. More fields, more integrations, more dashboards, more segments and more automations all create the appearance of maturity. In reality, unused capability is not maturity. It is operational debt with a license fee.
Webinar
Jun
16
From Legacy to Launch-Ready: How Gainbridge Made Its Website a Marketing-Led Growth Engine
Join in to learn how a D2C annuity brand gave marketing full website ownership — without slowing down or risking compliance.
Register
Webinar
Jun
18
The 5-Question CX Audit: Benchmark Your CX Operations for 2026
Built around insights from the 2026 CX Outsourcing Report, this live session puts the audit into practice.
Register
Webinar
Jun
30
How Modern Marketing Is Exposing the Limits of Legacy CMS
Why marketers are rethinking CMS workflows that slow publishing, personalization and campaign execution.
Register
Webinar
Jul
14
Content Leaders Collective: Proving Content’s Business Impact
Join us as top content leaders look beyond the buzzwords to share how they actually prove ROI and scale what works.
Register
Webinar
On demand
Content Strategy Leaders Live: Managing Risk, Compliance & AI in Financial Services
Learn how financial services leaders are modernizing content systems without disrupting trust, compliance or experience.
Watch Now
Webinar
On demand
From AI Investment to CX Results: What Enterprise Leaders Need to Know
Move beyond experiments. See how top enterprises scale AI for CX results.
Watch Now
Gartner’s CMO Spend and Strategy Survey exposed this problem before the current AI cycle hit full speed. According to coverage of the Gartner findings, marketers used only 33% of their martech stack’s capability in 2023, down from 42% in 2022 and 58% in 2020. At the same time, marketing technology accounted for 25.4% of the marketing budget. That is not a feature gap. It is a utilization failure.
This pattern should make AI leaders uncomfortable. If teams already use only one third of the tools they bought, adding AI will not magically produce better execution. It will add another layer that still requires decision logic, ownership and measurement. More capability does not fix a team that has not defined how capability changes behavior. Stop optimizing layers, fix the core.
The Last-Mile Problem AI Can't Solve Alone
Willis described the missing layer as a last-mile problem. “The biggest gap we have observed is getting the right information to the right person at the right time in a way that it drives the right decision,” he said. He also noted that much of the data and AI market focuses on integrations, pipelines, storage, retrieval and analysis, while less attention goes to application. That last word matters most.
Application is where the customer experience either improves or breaks. A data pipeline does not retain a customer. A predictive model does not repair a bad process. A dashboard does not resolve friction. Execution over interpretation.
Real Time Can Make the Wrong System Worse
Real time decisioning sounds like the natural next step for AI-powered customer experience. The system detects behavior, selects the next action and delivers the message at the right moment. The pitch is clean, fast and appealing. It also hides the operational risk.
Real time does not lower the bar. It raises it. Batch workflows give teams time to see conflicts, correct logic and prevent overlapping journeys. Real time systems reduce that cushion. Bad rules execute faster, and weak governance breaks in public.
Willis framed the requirement plainly.
“Real time requires tighter discipline behind desired outcomes and introduces a higher cost,” he said. He described these workloads as a deep investment in resources, process and infrastructure. That matters for every CRM leader being asked to make journeys more automated, predictive and immediate. Speed without ownership is not sophistication.
Where Effort Should Actually Go
BCG’s 10-20-70 rule for AI transformation offers a useful corrective to the platform-first mindset. The model directs roughly 10% of effort to algorithms, 20% to technology and data, and 70% to people, processes and operating model. That split is not flashy. It is practical.
Marketing teams see the stakes every day. A retention journey collides with a promotional campaign. A service recovery message overlaps with a sales push. A loyalty offer reaches someone whose last issue remains unresolved. Real time AI can manage those moments only when the business defines the decision logic behind them.
Decision Design Turns Signals Into Action
Decision design is the missing operating layer. It defines what signal matters, what decision follows, who owns the response and how success gets measured. It is not dashboard design. It is not journey mapping. It is not model tuning. Those disciplines show, sequence or predict, while decision design governs what the business does next.
Take email attrition as a practical example. A customer clicks one in three messages, then one in six, then none for 60 days. The reporting shows the decline, the engagement score drops and the automation platform detects the behavior. Many teams still treat the eventual unsubscribe as routine list hygiene. Customers leave because something broke.
Decision design treats the same signal differently. It defines the threshold before the unsubscribe happens. When a customer records no clicks across a defined number of sends or no engagement for a set window, the lifecycle team owns an intervention within seven days. The response can pause broad promotional sends, shift the customer into a value-led re-engagement path, request a preference update or move the customer into another channel.
That is the difference between watching and acting. The signal did not change, and the model did not need to become smarter. The organization designed a path that converted the signal into a decision, assigned ownership and measured the result. Consistency beats prediction when prediction has no path to action.
The Four Components of a Decision Path
For CRM and marketing teams, the decision path requires four components:
- Signal: The observed behavior that requires a decision.
- Decision: The action that changes because of that signal.
- Owner: The person or team responsible for action within a defined timeframe.
- Measure: The outcome that proves the decision worked.
This structure sounds simple because it is simple. It is also the work many teams avoid. They would rather tune the model, add another dashboard or buy another layer than define the operational decision. That avoidance explains why AI programs stall after promising starts.
AI Decision-Making FAQ
Editor's note: These are the key questions marketing, CRM and customer experience leaders are asking as AI adoption outpaces operational readiness.
Why are many AI initiatives failing to deliver business value?
Many organizations have invested in AI tools, dashboards and predictive models but have not redesigned workflows, ownership structures and decision-making processes around them. AI can identify opportunities and risks, but organizations still need a defined process for acting on those insights.
What is the AI execution gap?
The AI execution gap is the disconnect between generating insights and taking action. Organizations often deploy AI successfully but fail to define who owns the resulting decision, what action should occur and how success will be measured.
How to Convert AI Signals Into Action
The most productive AI question is not which model to use. The better question is which decision needs to improve. That shift forces CRM, marketing and martech teams to move from tool selection to operating design. It also exposes weak points that platforms tend to hide, including unclear goals, competing owners, inconsistent thresholds and success measures that do not match business outcomes.
Willis offered the starting point.
“Start with core objectives and measures,” he said. “Establishing and communicating core objectives drives everything else into alignment.” For marketing teams, the practical version is even sharper. Start with the customer behavior that must change, then define the internal decision that must change first.
This is where AI becomes valuable. It can detect friction earlier, summarize large volumes of feedback, classify intent and recommend interventions. It can help teams move faster once the operating model supports the decision. It cannot define the business outcome, assign decision rights or resolve process conflict without structure.
The contrarian point is not that AI is overhyped. The sharper point is that AI gets blamed for failures it did not create. Marketing already had a decision problem, CRM already had an execution problem and martech already had a utilization problem. AI entered the room and made those problems easier to see.
That is good news for teams willing to fix the system. The path forward does not require waiting for a better model. It requires building the decision path that should have existed before AI arrived. Define the signal, define the decision, define the owner and define the measure.
Then let AI accelerate a system worth accelerating.
Learn how you can join our contributor community.